From maine@altair.dfrf.nasa.gov Tue Feb 28 22:31:03 1995 Received: from altair.dfrf.nasa.gov by dkuug.dk with SMTP id AA06970 (5.65c8/IDA-1.4.4j for ); Tue, 28 Feb 1995 23:30:56 +0100 Received: by altair.dfrf.nasa.gov (5.0/SMI-SVR4) id AA07408; Tue, 28 Feb 1995 14:31:03 +0800 Date: Tue, 28 Feb 1995 14:31:03 +0800 From: maine@altair.dfrf.nasa.gov (Richard Maine) Message-Id: <9502282231.AA07408@altair.dfrf.nasa.gov> To: jwagener@amoco.com Cc: SC22WG5@dkuug.dk In-Reply-To: <199502282126.AA06150@dkuug.dk> (jwagener@amoco.com) Subject: Re: (SC22WG5.738) Content-Length: 1492 X-Charset: ASCII X-Char-Esc: 29 I like the idea of starting some real technical discussion of this interoperability question. I do regard it as a very important area. So, to try to address technical questions... Parts of Jerry's proposal that I agree with: 1. Restrict the current consideration to C, though leaving open the possibility of future expansion. This might tie down the problem to something narrow enough to actually get solved quickly enough to be useful. 2. Use some extension of the f90 interface block. This seems like an "obviously" good approach. Things I might argue with: 1. Jerry suggested that some people might prefer restricting the argument association to be with prespecified compatable types. I think I'd prefer this. Otherwise, we get essentially no argument checking. We'd be like PL/I, where you could freely pass a numeric argument to a character dummy, or vice-versa, and the compiler would do its best to do something. This is, to me, the trickiest area. It might not be obvious what the corresponding types were in all cases. For instance, does a C longint correspond to a default integer or some other integer kind? Assumptions about this might not be portable. Perhaps we could allow kind mismatch, but not type mismatch? 2. As long as we are talking about interoperability, email messages are more interoperable if there are line breaks other than between paragraphs . -- Richard Maine maine@altair.dfrf.nasa.gov